Millimeter-wave Passive Filtering Component Design

In recent years,with the rapid development of millimeter wave wireless communication technology and the urgent demand of high capacity data transmission rate,it is necessary to seek spectrum resources in a higher frequency band.China takes24.75-27.5 GHz and 37-42.5 GHz as the millimeter wave communication frequency band of 5G,which is rich in resources and can meet the requirements of broadband design.As one of the important components of millimeter wave wireless communication system,passive circuit design is always a research hotspot.At the same time,with the increase of frequency,passive circuit system design in the face of such problems as miniaturization,broadband,circuit design is becoming more and more complex,can optimize the number of arguments,optimization of difficulty and optimize the costs are increasing,the applications of machine learning techniques to optimize process in order to reduce computation cost,improve the efficiency of optimization is also one of the hotspots of research.In this paper,the power divider and filter of millimeter wave passive circuit are studied,and the machine learning-assisted optimization method is used to achieve rapid optimization.The main research results are as follows:Firstly,based on the substrate integrated coaxial line technology,the broadband SICL power divider of 16-way is designed for the 20-45 GHz band.Also,based on the 2-way SICL power divider,a dual-frequency filtering power divider is designed for24.25-27.5 GHz and 37-40 GHz bands by adding transmission lines and parallel short-circuit branch lines between output ports.Then,based on the substrate integrated waveguide technology,two bandpass SIW filters are designed for 24.75-27.5 GHz and 37-42.5 GHz respectively.According to the actual processing requirements,the position of the first metallized through hole is biased to achieve the same effect as the original metallized through hole with too small diameter.Finally,in the above design process,the machine learning-assisted optimization method is used to optimize the designed millimeter wave passive filtering component,and the correctness of the design is verified by the method of machining measurement.In order to facilitate testing,after designing and adding the transfer structure separately,the measured results are as follows: For the filtering power divider,the 10 d B bandwidth is 4.71 GHz(22.91-27.62 GHz,18.64%)and 4.4 GHz(36.11-40.51 GHz,11.49%),and in the passband,the insertion loss is less than 4.2 and 4.8 d B,respectively.For the26 GHz SIW bandpass filter,the 3d B bandwidth is 2.847 GHz(24.33-27.18 GHz,10.85%).In the passband,the return loss is greater than 19.1 d B,the insertion loss is less than 1 d B.For the 39 GHz SIW bandpass filter,the 3d B bandwidth is 9.70 GHz(35.45-45.15 GHz,24.07%).In the passband,the return loss is greater than 18.2 dB,the insertion loss is less than 1.5 dB.